This research article describes a study of the application of ARIMA modeling techniques to the Nigeria Gross Domestic Products between the periods 1980 and 2007, obtained from CBN Statistical Bulletin. For statistical analysis we have used graphical methods to display data distributions, Autocorrelation functions (ACF), Partial autocorrelation functions (PACF), Residuals and Forecasts, and differencing to check for stationarity. Our results are summarized as follows: The ARIMA (1, 2, 1) model was proposed for the data from the second differences which shows stability and invertibility, forecasts were made for future observations up to thirteen (13) years which shows an increasing trend over time, and the Akaike Information Criterion (AIC) and the adjusted multiple correlation coefficient (Adjusted R-Square) provided a good summary of the total variability explained by the chosen fitted model. In conclusion, the article demonstrated the use of autocorrelations and partial autocorrelations for identifying an ARIMA (p, d, q) model for the series using Box and Jenkins approach.
Time Series, CBN, ARIMA, Modeling, GDP, Stochastic, Nonstationary, Stationarity, Stability, Invertibility, Identification, ACF, PACF, Differencing, Diagnostics, Estimation, Residuals, Forecast, AIC, Adjusted R-Square
